1 | /* Vector optimized 32/64 bit S/390 version of wcschr. |
2 | Copyright (C) 2015-2022 Free Software Foundation, Inc. |
3 | This file is part of the GNU C Library. |
4 | |
5 | The GNU C Library is free software; you can redistribute it and/or |
6 | modify it under the terms of the GNU Lesser General Public |
7 | License as published by the Free Software Foundation; either |
8 | version 2.1 of the License, or (at your option) any later version. |
9 | |
10 | The GNU C Library is distributed in the hope that it will be useful, |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
13 | Lesser General Public License for more details. |
14 | |
15 | You should have received a copy of the GNU Lesser General Public |
16 | License along with the GNU C Library; if not, see |
17 | <https://www.gnu.org/licenses/>. */ |
18 | |
19 | #include <ifunc-wcschr.h> |
20 | #if HAVE_WCSCHR_Z13 |
21 | |
22 | # include "sysdep.h" |
23 | # include "asm-syntax.h" |
24 | |
25 | .text |
26 | |
27 | /* wchar_t *wcschr (const wchar_t *s, wchar_t c) |
28 | Locate character in string. |
29 | |
30 | Register usage: |
31 | -r1=tmp |
32 | -r2=s |
33 | -r3=c |
34 | -r4=tmp |
35 | -r5=current_len |
36 | -v16=part of s |
37 | -v17=index of unequal |
38 | -v18=replicated c |
39 | */ |
40 | ENTRY(WCSCHR_Z13) |
41 | .machine "z13" |
42 | .machinemode "zarch_nohighgprs" |
43 | |
44 | vlbb %v16,0(%r2),6 /* Load s until next 4k-byte boundary. */ |
45 | lcbb %r1,0(%r2),6 /* Get bytes to 4k-byte boundary or 16. */ |
46 | |
47 | tmll %r2,3 /* Test if s is 4-byte aligned? */ |
48 | jne .Lfallback /* And use common-code variant if not. */ |
49 | |
50 | lghi %r5,0 /* current_len = 0. */ |
51 | |
52 | vlvgf %v18,%r3,0 /* Generate vector which elements are all c. */ |
53 | vrepf %v18,%v18,0 |
54 | |
55 | vfeezfs %v16,%v16,%v18 /* Find element equal with zero search. */ |
56 | vlgvb %r4,%v16,7 /* Load byte index of character or zero. */ |
57 | clrjl %r4,%r1,.Lfound /* Return if c/zero is in loaded bytes. */ |
58 | |
59 | /* Align s to 16 byte. */ |
60 | risbgn %r4,%r2,60,128+63,0 /* %r3 = bits 60-63 of %r2 'and' 15. */ |
61 | lghi %r5,16 /* current_len = 16. */ |
62 | slr %r5,%r4 /* Compute bytes to 16bytes boundary. */ |
63 | |
64 | /* Find c/zero in 16byte aligned loop */ |
65 | .Lloop: |
66 | vl %v16,0(%r5,%r2) /* Load s. */ |
67 | vfeezfs %v16,%v16,%v18 /* Find element equal with zero search. */ |
68 | jno .Lfound /* Found c/zero (cc=0|1|2). */ |
69 | vl %v16,16(%r5,%r2) |
70 | vfeezfs %v16,%v16,%v18 |
71 | jno .Lfound16 |
72 | vl %v16,32(%r5,%r2) |
73 | vfeezfs %v16,%v16,%v18 |
74 | jno .Lfound32 |
75 | vl %v16,48(%r5,%r2) |
76 | vfeezfs %v16,%v16,%v18 |
77 | jno .Lfound48 |
78 | |
79 | aghi %r5,64 |
80 | j .Lloop /* No character and no zero -> loop. */ |
81 | |
82 | .Lfound48: |
83 | la %r5,16(%r5) /* Use la since aghi would clobber cc. */ |
84 | .Lfound32: |
85 | la %r5,16(%r5) |
86 | .Lfound16: |
87 | la %r5,16(%r5) |
88 | .Lfound: |
89 | je .Lzero /* Found zero, but no c before that zero. */ |
90 | |
91 | .Lcharacter: |
92 | vlgvb %r4,%v16,7 /* Load byte index of character. */ |
93 | algr %r5,%r4 |
94 | la %r2,0(%r5,%r2) /* Return pointer to character. */ |
95 | br %r14 |
96 | |
97 | .Lzero: |
98 | clije %r3,0,.Lcharacter /* Found zero and c is zero. */ |
99 | lghi %r2,0 /* Return null if character not found. */ |
100 | br %r14 |
101 | .Lfallback: |
102 | jg WCSCHR_C |
103 | END(WCSCHR_Z13) |
104 | |
105 | # if ! HAVE_WCSCHR_IFUNC |
106 | strong_alias (WCSCHR_Z13, __wcschr) |
107 | weak_alias (__wcschr, wcschr) |
108 | # endif |
109 | |
110 | # if defined HAVE_S390_MIN_Z13_ZARCH_ASM_SUPPORT \ |
111 | && defined SHARED && IS_IN (libc) |
112 | strong_alias (WCSCHR_Z13, __GI___wcschr) |
113 | weak_alias (WCSCHR_Z13, __GI_wcschr) |
114 | # endif |
115 | #endif |
116 | |